A growing body of research in gerontology links social interaction in old age to multiple indicators of psychological well-being and physical health. As a whole this research provides compelling evidence for the contention that social relationships are fundamental to the well-being of an aging population. Findings from recent studies of older adults, however, prompt reconsideration of the notion that social interaction is uniformly positive. In this application, I propose a model of socioemotional aging that assumes that social interaction in old age differs from social interaction earlier in life; and, further, that age differences are accounted for by both psychological differences in older adults and interpersonal changes that can function to limit affective gain derived from social interaction. In this application, a four year program of research is described aimed at the identification of the costs and the benefits -- both social and emotional -- of social interaction for the older person and his or her social partners. Four studies are proposed that collectively employ multiple research methods, including interviews with older and younger subjects about their preferences for and selection of social partners; videotaped emotion induction procedures with an examination age-differences in the duration and intensity of expressed emotion and ratings of emotion expression as perceived by others; direct observation of same-aged and mixed-aged dyadic interaction in the laboratory; and time-sampling of activities and affect in naturalistic settings. Despite continued disagreement about the meaning of lowered rates of social interaction common in old age, the basic finding of an ontogenetic reduction in rates of interaction remains undisputed. As we face a rapidly aging population, it is more essential than ever to understand the meaning and the mechanisms responsible for these reductions.